Service guarantees/congestion control in high speed networks

Abstract

Congestion control is implemented at the resources that actually suffer from congestion and, moreover the congestion control is performed within the network as well as at its access points. In a particular embodiment, congestion control is continuously performed at each resource. Use of each resource is controlled according to a head-of-the-line weighted round-robin service discipline thereby assuring fairness and full use of bandwidth for all active channels. Also, in order to control conditions of congestion, the queue of data elements waiting for service at each resource is regulated by a global congestion indicator for the entire queue. The data elements of each channel that are waiting for service are further regulated in response to specific local congestion indicators that are individualized for each channel. The global and local congestion indicators are used to guarantee predetermined levels of service for each channel. The values employed for the head-of-the-line weighted round-robin service discipline are developed as a function of at least the contracted-for bandwidth for each channel. The developed values are replicated for each resource subject to congestion control.